Abstract

Background Osteosarcoma is the most common malignant bone disease. The current clinical treatment is difficult to deal with the high heterogeneity within tumors and the cellular dynamic remodeling during the metastasis process of osteosarcoma. Methods Cell annotation was performed using the SingleR algorithm in combination with specific marker genes. Further subcluster analysis was conducted on osteoblasts, myeloid cells, fibroblasts, malignant cells and T/NK cells. Cell differentiation trajectories were constructed using Monocle2 and scTour for pseudotime analysis. The spatial transcriptome data was deconvolved using the CARD algorithm to verify the spatial distribution of cells. Results Compared with primary tumors, the microenvironment of metastatic tumors undergoes significant remodeling, manifested as a significant increase in the proportions of fibroblasts, mast cells and T/NK cells, while the proportions of osteoblasts and myeloid cells decrease. Spatial bifurcation pattern revealed a divergent evolutionary path from subtype 2 to other subtypes as the metastatic state. It identified YY1 in subtype2 as a spatial regulator of IER5L in the osteosarcoma microenvironment. The study also observed the reprogramming of immune cells and stromal cells in metastatic foci. In particular, the differentiation trajectory of TAMs (SLC40A1 → MT1G → CXCL10) suggests that macrophages undergo functional transformation during tumor progression under an immunosuppressive microenvironment. Conclusion It is the most forward time for the large-scale integration of multi-center single-cell data and spatial transcriptomics profiles covering the primary, recurrent and metastatic stages, systematically depicting the dynamic changes in cell composition during the progression of osteosarcoma.